<?xml version="1.0" encoding="utf-8"?>
<TEI xmlns="http://www.tei-c.org/ns/1.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:hal="http://hal.archives-ouvertes.fr/" xmlns:gml="http://www.opengis.net/gml/3.3/" xmlns:gmlce="http://www.opengis.net/gml/3.3/ce" version="1.1" xsi:schemaLocation="http://www.tei-c.org/ns/1.0 http://api.archives-ouvertes.fr/documents/aofr-sword.xsd">
  <teiHeader>
    <fileDesc>
      <titleStmt>
        <title>HAL TEI export of hal-03382553</title>
      </titleStmt>
      <publicationStmt>
        <distributor>CCSD</distributor>
        <availability status="restricted">
          <licence target="https://creativecommons.org/publicdomain/zero/1.0/">CC0 1.0 - Universal</licence>
        </availability>
        <date when="2026-05-17T18:25:36+02:00"/>
      </publicationStmt>
      <sourceDesc>
        <p part="N">HAL API Platform</p>
      </sourceDesc>
    </fileDesc>
  </teiHeader>
  <text>
    <body>
      <listBibl>
        <biblFull>
          <titleStmt>
            <title xml:lang="en">On the Universality of Graph Neural Networks on Large Random Graphs</title>
            <author role="aut">
              <persName>
                <forename type="first">Nicolas</forename>
                <surname>Keriven</surname>
              </persName>
              <email type="md5">fb84266a29a185392c998b2b5317f7e7</email>
              <email type="domain">cnrs.fr</email>
              <ptr type="url" target="https://nkeriven.github.io"/>
              <idno type="idhal" notation="string">nicolas-keriven</idno>
              <idno type="idhal" notation="numeric">4423</idno>
              <idno type="halauthorid" notation="string">25971-4423</idno>
              <idno type="IDREF">https://www.idref.fr/223562270</idno>
              <idno type="ORCID">https://orcid.org/0000-0002-3846-8763</idno>
              <idno type="VIAF">https://viaf.org/viaf/120151836552920401906</idno>
              <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=aNYFO8sAAAAJ</idno>
              <affiliation ref="#struct-1043350"/>
              <affiliation ref="#struct-441569"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Alberto</forename>
                <surname>Bietti</surname>
              </persName>
              <idno type="halauthorid">860293-0</idno>
              <affiliation ref="#struct-1079527"/>
            </author>
            <author role="aut">
              <persName>
                <forename type="first">Samuel</forename>
                <surname>Vaiter</surname>
              </persName>
              <email type="md5">5ac6602d81d93dccbee3ff5e790f0e84</email>
              <email type="domain">math.cnrs.fr</email>
              <idno type="idhal" notation="string">samuel-vaiter</idno>
              <idno type="idhal" notation="numeric">1995</idno>
              <idno type="halauthorid" notation="string">38917-1995</idno>
              <idno type="IDREF">https://www.idref.fr/182993116</idno>
              <idno type="ORCID">https://orcid.org/0000-0002-4077-708X</idno>
              <idno type="VIAF">https://viaf.org/viaf/313572753</idno>
              <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=HkXkm7IAAAAJ</idno>
              <affiliation ref="#struct-300270"/>
            </author>
            <editor role="depositor">
              <persName>
                <forename>Nicolas</forename>
                <surname>Keriven</surname>
              </persName>
              <email type="md5">fb84266a29a185392c998b2b5317f7e7</email>
              <email type="domain">cnrs.fr</email>
            </editor>
          </titleStmt>
          <editionStmt>
            <edition n="v1" type="current">
              <date type="whenSubmitted">2021-10-18 11:39:39</date>
              <date type="whenWritten">2021-05</date>
              <date type="whenModified">2026-01-07 11:02:02</date>
              <date type="whenReleased">2021-10-18 11:39:39</date>
              <date type="whenProduced">2021-12-06</date>
              <ref type="externalLink" target="http://arxiv.org/pdf/2105.13099"/>
            </edition>
            <respStmt>
              <resp>contributor</resp>
              <name key="317025">
                <persName>
                  <forename>Nicolas</forename>
                  <surname>Keriven</surname>
                </persName>
                <email type="md5">fb84266a29a185392c998b2b5317f7e7</email>
                <email type="domain">cnrs.fr</email>
              </name>
            </respStmt>
          </editionStmt>
          <publicationStmt>
            <distributor>CCSD</distributor>
            <idno type="halId">hal-03382553</idno>
            <idno type="halUri">https://hal.science/hal-03382553</idno>
            <idno type="halBibtex">keriven:hal-03382553</idno>
            <idno type="halRefHtml">&lt;i&gt;NeurIPS 2021 - 35th Conference on Neural Information Processing Systems&lt;/i&gt;, Dec 2021, Virtual, Canada</idno>
            <idno type="halRef">NeurIPS 2021 - 35th Conference on Neural Information Processing Systems, Dec 2021, Virtual, Canada</idno>
            <availability status="restricted"/>
          </publicationStmt>
          <seriesStmt>
            <idno type="stamp" n="UNIV-BOURGOGNE">Université Bourgogne Europe</idno>
            <idno type="stamp" n="UGA">HAL Grenoble Alpes</idno>
            <idno type="stamp" n="CNRS">CNRS - Centre national de la recherche scientifique</idno>
            <idno type="stamp" n="INPG">Institut polytechnique de Grenoble</idno>
            <idno type="stamp" n="GIPSA">Grenoble Images Parole Signal Automatique</idno>
            <idno type="stamp" n="DIEUDONNE">Laboratoire Jean-Alexandre Dieudonné</idno>
            <idno type="stamp" n="DIEUDONNE-PS" corresp="DIEUDONNE">Laboratoire J.A. Dieudonné - UMR 7351 - Probabilités et Statistiques</idno>
            <idno type="stamp" n="GIPSA-GAIA" corresp="GIPSA">Département Géométrie, Apprentissage, Information et Algorithmes</idno>
            <idno type="stamp" n="UGA-EPE">Université Grenoble Alpes [2020-*]</idno>
            <idno type="stamp" n="TEST-UGA">TEST-UGA</idno>
          </seriesStmt>
          <notesStmt>
            <note type="audience" n="2">International</note>
            <note type="invited" n="0">No</note>
            <note type="popular" n="0">No</note>
            <note type="peer" n="1">Yes</note>
            <note type="proceedings" n="1">Yes</note>
          </notesStmt>
          <sourceDesc>
            <biblStruct>
              <analytic>
                <title xml:lang="en">On the Universality of Graph Neural Networks on Large Random Graphs</title>
                <author role="aut">
                  <persName>
                    <forename type="first">Nicolas</forename>
                    <surname>Keriven</surname>
                  </persName>
                  <email type="md5">fb84266a29a185392c998b2b5317f7e7</email>
                  <email type="domain">cnrs.fr</email>
                  <ptr type="url" target="https://nkeriven.github.io"/>
                  <idno type="idhal" notation="string">nicolas-keriven</idno>
                  <idno type="idhal" notation="numeric">4423</idno>
                  <idno type="halauthorid" notation="string">25971-4423</idno>
                  <idno type="IDREF">https://www.idref.fr/223562270</idno>
                  <idno type="ORCID">https://orcid.org/0000-0002-3846-8763</idno>
                  <idno type="VIAF">https://viaf.org/viaf/120151836552920401906</idno>
                  <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=aNYFO8sAAAAJ</idno>
                  <affiliation ref="#struct-1043350"/>
                  <affiliation ref="#struct-441569"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Alberto</forename>
                    <surname>Bietti</surname>
                  </persName>
                  <idno type="halauthorid">860293-0</idno>
                  <affiliation ref="#struct-1079527"/>
                </author>
                <author role="aut">
                  <persName>
                    <forename type="first">Samuel</forename>
                    <surname>Vaiter</surname>
                  </persName>
                  <email type="md5">5ac6602d81d93dccbee3ff5e790f0e84</email>
                  <email type="domain">math.cnrs.fr</email>
                  <idno type="idhal" notation="string">samuel-vaiter</idno>
                  <idno type="idhal" notation="numeric">1995</idno>
                  <idno type="halauthorid" notation="string">38917-1995</idno>
                  <idno type="IDREF">https://www.idref.fr/182993116</idno>
                  <idno type="ORCID">https://orcid.org/0000-0002-4077-708X</idno>
                  <idno type="VIAF">https://viaf.org/viaf/313572753</idno>
                  <idno type="GOOGLE SCHOLAR">https://scholar.google.fr/citations?user=HkXkm7IAAAAJ</idno>
                  <affiliation ref="#struct-300270"/>
                </author>
              </analytic>
              <monogr>
                <meeting>
                  <title>NeurIPS 2021 - 35th Conference on Neural Information Processing Systems</title>
                  <date type="start">2021-12-06</date>
                  <date type="end">2021-12-14</date>
                  <settlement>Virtual</settlement>
                  <country key="CA">Canada</country>
                </meeting>
                <imprint>
                  <date type="datePub">2021-12</date>
                </imprint>
              </monogr>
              <idno type="arxiv">2105.13099</idno>
              <ref type="publisher">https://proceedings.neurips.cc/paper/2021</ref>
            </biblStruct>
          </sourceDesc>
          <profileDesc>
            <langUsage>
              <language ident="en">English</language>
            </langUsage>
            <textClass>
              <classCode scheme="halDomain" n="stat.ml">Statistics [stat]/Machine Learning [stat.ML]</classCode>
              <classCode scheme="halDomain" n="math.math-st">Mathematics [math]/Statistics [math.ST]</classCode>
              <classCode scheme="halTypology" n="COMM">Conference papers</classCode>
              <classCode scheme="halOldTypology" n="COMM">Conference papers</classCode>
              <classCode scheme="halTreeTypology" n="COMM">Conference papers</classCode>
            </textClass>
            <abstract xml:lang="en">
              <p>We study the approximation power of Graph Neural Networks (GNNs) on latent position random graphs. In the large graph limit, GNNs are known to converge to certain "continuous" models known as c-GNNs, which directly enables a study of their approximation power on random graph models. In the absence of input node features however, just as GNNs are limited by the Weisfeiler-Lehman isomorphism test, c-GNNs will be severely limited on simple random graph models. For instance, they will fail to distinguish the communities of a well-separated Stochastic Block Model (SBM) with constant degree function. Thus, we consider recently proposed architectures that augment GNNs with unique node identifiers, referred to as Structural GNNs here (SGNNs). We study the convergence of SGNNs to their continuous counterpart (c-SGNNs) in the large random graph limit, under new conditions on the node identifiers. We then show that c-SGNNs are strictly more powerful than c-GNNs in the continuous limit, and prove their universality on several random graph models of interest, including most SBMs and a large class of random geometric graphs. Our results cover both permutation-invariant and permutation-equivariant architectures.</p>
            </abstract>
          </profileDesc>
        </biblFull>
      </listBibl>
    </body>
    <back>
      <listOrg type="structures">
        <org type="department" xml:id="struct-1043350" status="OLD">
          <orgName>GIPSA Pôle Géométrie, Apprentissage, Information et Algorithmes</orgName>
          <orgName type="acronym">GIPSA-GAIA</orgName>
          <date type="start">2020-01-01</date>
          <date type="end">2025-12-31</date>
          <desc>
            <address>
              <addrLine>GIPSA-lab11 rue des Mathématiques, Grenoble Campus BP46F-38402 SAINT MARTIN D'HERES CEDEX</addrLine>
              <country key="FR"/>
            </address>
          </desc>
          <listRelation>
            <relation active="#struct-1043333" type="direct"/>
            <relation name="UMR5216" active="#struct-441569" type="indirect"/>
            <relation active="#struct-1042703" type="indirect"/>
            <relation active="#struct-1043329" type="indirect"/>
          </listRelation>
        </org>
        <org type="regroupinstitution" xml:id="struct-441569" status="VALID">
          <idno type="IdRef">02636817X</idno>
          <idno type="ISNI">0000000122597504</idno>
          <idno type="ROR">https://ror.org/02feahw73</idno>
          <orgName>Centre National de la Recherche Scientifique</orgName>
          <orgName type="acronym">CNRS</orgName>
          <date type="start">1939-10-19</date>
          <desc>
            <address>
              <country key="FR"/>
            </address>
            <ref type="url">https://www.cnrs.fr/</ref>
          </desc>
        </org>
        <org type="laboratory" xml:id="struct-1079527" status="VALID">
          <orgName>Center for Data Science [NYU]</orgName>
          <orgName type="acronym">CDS</orgName>
          <date type="start">2013-01-01</date>
          <desc>
            <address>
              <addrLine>New York University, CDS,  7th floor, 60 5th Ave, New York, NY, 1001</addrLine>
              <country key="US"/>
            </address>
            <ref type="url">https://cds.nyu.edu/</ref>
          </desc>
          <listRelation>
            <relation active="#struct-300459" type="direct"/>
            <relation active="#struct-566779" type="indirect"/>
          </listRelation>
        </org>
        <org type="institution" xml:id="struct-300270" status="OLD">
          <idno type="IdRef">02819005X</idno>
          <idno type="ISNI">0000000122989313</idno>
          <idno type="ROR">https://ror.org/03k1bsr36</idno>
          <orgName>Université de Bourgogne</orgName>
          <orgName type="acronym">UB</orgName>
          <date type="end">2024-12-31</date>
          <desc>
            <address>
              <addrLine>Maison de l'université - Esplanade Érasme - BP 27877 - 21078 Dijon cedex</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.u-bourgogne.fr/</ref>
          </desc>
        </org>
        <org type="laboratory" xml:id="struct-1043333" status="VALID">
          <idno type="IdRef">137251122</idno>
          <idno type="ISNI">0000000118823396</idno>
          <idno type="RNSR">200711885T</idno>
          <idno type="ROR">https://ror.org/02wrme198</idno>
          <orgName>Grenoble Images Parole Signal Automatique</orgName>
          <orgName type="acronym">GIPSA-lab</orgName>
          <date type="start">2020-01-01</date>
          <desc>
            <address>
              <addrLine>GIPSA-lab 11 rue des Mathématiques, Grenoble Campus BP46F-38402 SAINT MARTIN D'HERES CEDEX</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.gipsa-lab.inpg.fr/</ref>
          </desc>
          <listRelation>
            <relation name="UMR5216" active="#struct-441569" type="direct"/>
            <relation active="#struct-1042703" type="direct"/>
            <relation active="#struct-1043329" type="direct"/>
          </listRelation>
        </org>
        <org type="regroupinstitution" xml:id="struct-1042703" status="VALID">
          <idno type="IdRef">240648315</idno>
          <idno type="ROR">https://ror.org/02rx3b187</idno>
          <orgName>Université Grenoble Alpes</orgName>
          <orgName type="acronym">UGA</orgName>
          <date type="start">2020-01-01</date>
          <desc>
            <address>
              <addrLine>Adresse CS 40700 - 38058 Grenoble cedex</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.univ-grenoble-alpes.fr</ref>
          </desc>
        </org>
        <org type="institution" xml:id="struct-1043329" status="VALID">
          <idno type="IdRef">026388804</idno>
          <idno type="ROR">https://ror.org/05sbt2524</idno>
          <orgName>Institut polytechnique de Grenoble - Grenoble Institute of Technology</orgName>
          <orgName type="acronym">Grenoble INP</orgName>
          <date type="start">2020-01-01</date>
          <desc>
            <address>
              <addrLine>46 avenue Félix Viallet 38031 Grenoble Cedex 1</addrLine>
              <country key="FR"/>
            </address>
            <ref type="url">http://www.grenoble-inp.fr/</ref>
          </desc>
          <listRelation>
            <relation active="#struct-1042703" type="direct"/>
          </listRelation>
        </org>
        <org type="institution" xml:id="struct-300459" status="VALID">
          <idno type="ISNI">0000 0004 1936 8753</idno>
          <idno type="ROR">https://ror.org/0190ak572</idno>
          <orgName>New York University [New York]</orgName>
          <orgName type="acronym">NYU</orgName>
          <desc>
            <address>
              <addrLine>70 Washington Square South, New York, NY 10012</addrLine>
              <country key="US"/>
            </address>
            <ref type="url">http://www.nyu.edu/</ref>
          </desc>
          <listRelation>
            <relation active="#struct-566779" type="direct"/>
          </listRelation>
        </org>
        <org type="regroupinstitution" xml:id="struct-566779" status="VALID">
          <orgName>NYU System</orgName>
          <orgName type="acronym">NYU</orgName>
          <desc>
            <address>
              <country key="US"/>
            </address>
            <ref type="url">http://www.nyu.edu/</ref>
          </desc>
        </org>
      </listOrg>
    </back>
  </text>
</TEI>